¹³C Enrichment of Carbons 2 and 8 of Purine by Folate-Dependent Reactions After [¹³C]Formate and [2-¹³C]Glycine Dosing in Adult Humans

Abstract

The 10-formyl moiety of 10-formyltetrahydrofolate is the source of carbons at the positions 8 (C₈) and 2 (C₂) of the purine ring, originating from formate and a few amino acids. Uric acid is the final catabolic product of purines. In adult humans, we independently measured the ¹³C enrichment of the C₂ and C₈ positions of urinary uric acid after an oral dose of [¹³C]sodium formate and that of the C₂ and C₈ plus C₅ positions after [2-¹³C]glycine. A liquid-chromatography mass-spectrometric method was used to measure the ¹³C enrichment of uric acid in urine which was collected for 3 - 4 days. Purine catabolism to uric acid does not alter the positions of carbons in the ring. After the formate dose, the ¹³C-enrichment at C₂ was greater that at C₈, and a circadian rhythm was observed in the enrichment at C₂. After the glycine dose, the C₈ plus C₅ positions were enriched, whereas no significant enrichment at C₂ was found. These ¹³C enrichment patterns are not consistent with previous accepted metabolism. To our knowledge, this is the first study to investigate ¹³C enrichment from formate and glycine independently into the C₂ and C₈ positions of purine in the same subjects. Possible mechanisms explaining our findings are discussed. Oral [¹³C]formate or [2-¹³C]glycine dosing and urine collection can be used to study purine biosynthesis in humans.

1. Introduction

Purine de novo nucleotide biosynthesis is a fundamental process producing building blocks of DNA and RNA. Glycinamide ribotide (GAR) and aminoimidazolecarboxamide ribotide (AICAR) transformylases utilize folate coenzymes to introduce carbons 8 (C₈ in red, Fig. 1) and 2 (C₂ in blue, Fig. 1) into the purine ring, respectively [1].

Fig.1.

The origin of C₂, C₄ C₅, C₈ and N₇ atoms of the purine ring from formate and glycine. The C of formate may be incorporated into C₂ and C₈ via the 10-HCO-H₄folate synthetase, and GAR and AICAR transformylases. The first and second C and N of glycine are incorporated directly to C₄, C₅ and N₇, respectively (in broken lines). The second C of glycine may be incorporated into C₂ and C₈ via GCS and GAR and AICAR transformylases. All carbons potentially incorporated into C₂ and C₈ are in green, and C₂ and C₈ are in blue and red, respectively, which are matched with those of bars in Figure 2.

Formate is one source of the formyl C (green, Fig. 1) of 10-formyltetrahydrofolate (10-HCO-H₄folate), which is a substrate for these two enzymes. 10-HCO-H₄folate synthetase forms this substrate from tetrahydrofolate (H₄folate), formate and ATP [2]. Under normal conditions, human plasma formate concentrations range 20 - 250 μM, which is about 50% of those in erythrocytes, and formate is derived from many sources [3, 4]. In animals, ¹⁴C-labeled formate given in vivo is predominantly found at C₂ and C₈ of purines, whereas only a small amount appears in the other positions [5-10], and the ¹⁴C₂/¹⁴C₈ ratio was reported to be about 1.0 [6-8]. However, this ratio in humans has never been documented.

The second C of glycine (green, Fig. 1) is also potentially incorporated into C₂ and C₈ of purines by folate-dependent reactions and into C₅ by folate-independent metabolism. Glycine in the presence of H₄folate is metabolized to 5,10-methylenetetrahydrofolate (5,10-CH₂-H₄folate), CO₂ and NH₃ by the glycine-cleavage system (GCS) [11]. The second C of glycine (green, Fig. 1), now the methylene C of 5,10-CH₂-H₄folate, can be converted to 10-HCO-H₄folate by 5,10-CH₂-H₄folate dehydrogenase and 5,10-methenyltetrahydrofolate (5,10-CH=H₄folate) cyclohydrolase [2].

Uric acid is the final catabolite of purines in humans, and the catabolic process does not alter the C positions of purines [1]. In mammals, C₂ and C₈ of purines are not more labile to simple isotope exchange in vivo than the other Cs of purines [12, 13]. The peak in labeled urinary uric acid occurs 1 - 3 days after a dose of labeled formate or glycine in humans [14-19], representing catabolism of newly synthesized purines [14]. We measured the ¹³C enrichment independently at C₂ and C₈ of uric acid by expanding the in vivo formate pool with an oral [¹³C]formate dose using a liquid-chromatography mass spectrometry (LC/MS/MS) method [20]. We also measured the ¹³C enrichment at C₂ and C₈ plus C₅ after expanding the glycine pool with [2-¹³C]glycine.

2. Methods

2.1. Human study

The study was approved by the Institutional Review Board for Human Use at the University of Alabama at Birmingham. Three healthy adult males collected urine at each void for 72 - 96 hours after an oral dose of 14.5 mmol [¹³C]sodium formate (1.0 g, ¹³C 99%, Cambridge Isotope Laboratories, Andover, MA) with 100 mL of water at about noon. The volume of each void was measured and urine samples were stored at -80°C until analysis. In addition, as baseline experiments, similar urine collections for 72 hour were performed without a [¹³C]sodium formate dose (once in subject B and twice in subjects A and C). With a minimum of a 12-month interval, the identical procedure was repeated in two subjects using 41.7 mmol of [2-¹³C]glycine (2.5 g, ¹³C 99%, Cambridge Isotope Lab.). Subjects were asked to maintain regular life style including their diet and physical activity during the study.

2.2. Measurement of ¹³C enrichment

Independent ¹³C enrichment from [¹³C]formate into the C₂ and C₈ positions of uric acid was measured by the LC/MS/MS method [20] followed by the calculation as described below. The area-under-the-curve (AUC) of the extracted ion chromatograms (XIC) for m/z 168 → 124 and 168 → 125 and 169 → 125 represents the amount of ¹³C at C₂, C₈ and both C₂ and C₈, respectively. The AUC of XIC for m/z 167 → 124 represents uric acid containing only ¹²C, ¹H, ¹⁴N and ¹⁶O. The percentage of ¹³C at C₂(% ¹³C₂) and C₈ (% ¹³C₈) in total uric acid was estimated using the following formulae: % ¹³C₂ = (AUC for XIC m/z 168 → 124) ÷ (AUC for XIC m/z 167 → 124 + AUC for XIC m/z 169 → 125) × 100; and % ¹³C₈ = (AUC for XIC m/z 168 → 125) ÷ (AUC for XIC m/z 167 → 124 + AUC for XIC m/z 169 → 125) × 100.

Similar calculations were made to estimate the % ¹³C after a [2-¹³C]glycine dose. Although the LS/MS/MS method allows to measure % ¹³C at C₂ cleanly, % ¹³C at C₈ also includes that at C₅[20]. The amount of ¹³C at both C₂ and C₈ is included in the denominator because it is unlikely that both positions would be simultaneously enriched by a [¹³C]formate or [2-¹³C]glycine dose. The ¹³C enrichment at C₂ and C₈(plus C₅) following [¹³C]formate or [2-¹³C]glycine dosing was calculated for each void by subtracting baseline % ¹³C₂ and % ¹³C₈ values that were paired for the subject and time of d when void was collected from the values obtained after the dose. The values of % ¹³C₂ and % ¹³C₈ from baseline experiments were subtracted from each other in two subjects (A and C) to measure the variability of baseline values. It is important to note that the measurements of baseline values are essential for such a study, since life style or dietary habit can affect the ¹³C enrichment of human samples [21, 22].

2.3. Statistical analysis

The Wilcoxon paired-sample test was used to detect a significant difference from 0 in mean ¹³C enrichment from [¹³C]formate for each day, where more than eight voids were obtained. For subject C with fewer than eight voids per day, the data from three d were combined [23]. The same test was used to detect the ¹³C enrichment from [¹³C]glycine where data of three days were combined. The runs above-and below-the-median test (“runs test”) was performed to detect rhythmicity in subjects with greater than 25 voids. Significantly fewer than expected runs above- and below-the-median indicate a non-random temporal distribution of the data, suggesting a rhythmic pattern. This runs test avoids having to force fit the data to a cosine function. The details of the principle of runs test is presented by Sokal and Rohlf [23].

3. Results

3.1. [¹³C]Formate dose

The ¹³C enrichment after a [¹³C]formate dose did not significantly correlate with the amount of uric acid excreted in each void, uric acid concentration or urine volume. The¹³C enrichment at C₂ (blue columns) and C₈ (red columns) from subjects A and B is shown in Fig. 2. The peak % ¹³C enrichment from [¹³C]formate at C₂ was 0.74 - 5.7% and that at C₈ was 0.08 - 0.24%, and mean % ¹³C enrichment at C₂ was significantly greater than 0 in all three subjects (Table 1, Fig. 2). Mean % ¹³C enrichment at C₈ was significant greater than 0 in subjects B and C. Lower mean % ¹³C enrichments at C₈ were generally found compared to C₂. Thus, contrary to our expectation, the ¹³C enrichment ratio of C₂/C₈ was far from 1.0, and median C_2/C₈ ratios were 6.6, 6.5 and 3.0 for subjects A and B (obtained from the data presented in Fig. 2) and C (data not presented), respectively. Only positive ratios were used to determine the median. We observed that only 11 of 66 voids, where the C₂/C₈ ratios were positive, fell in the range 0.5 - 2.0, which could be generously considered close to 1.0.

Fig. 2.

The % ¹³C enrichment at C₂ and C₈ or C₈ plus C₅ of uric acid in each void after [¹³C]formate or [2-¹³C]glycine dosing. Bars represent ¹³C enrichment at C₂ (blue) and C₈ or C₈ plus C₅ (red) in voids collected for 72 - 96 hours after [¹³C] formate or [2-¹³C]glycine dosing in subjects A and B. The x-axis is time after dosing; however, the intervals between voids were different. Therefore, we did not show specify time of the void. The numbers next to the bars are % enrichments that were out of range on the y-axis.

Table 1.

Mean (± SD)% ¹³C enrichment at the C₂ and C₈ positions of uric acid after [¹³C]formate dose and at the C₂ and C₈ plus C₅ positions of uric acid after a [2-¹³C]glycine dose

Subject	Dose (mmol)	Day 1	Day 2	Day 3	Day 1	Day 2	Day 3
[13C]formate		% Enrichment at C2 (number of voids analyzed)			% Enrichment at C8 (number of voids analyzed)
A	14.5	0.58±0.25a (9)	0.42±0.09a (12)	0.47±0.19a (10)	0.00±0.06 (9)	0.03±0.04 (12)	0.02±0.04 (10)
B	14.5	0.84±0.78a (10)	0.51±0.78a (10)	0.40±0.73a (12)	0.04±0.03a (10)	0.04±0.03a (10)	0.05±0.03a (12)
A	0	-0.06±0.05 (8)	0.02±0.04 (9)	0.03±0.06 (9)	-0.01±0.09 (8)	0.04±0.08 (9)	0.04±0.08 (9)
C	14.5	0.43 ± 0.18a (16)			0.16 ± 0.04a (16)
C	0	0.03 ± 0.16 (16)			0.01 ± 0.09 (16)

[2-13C]glycine		% Enrichment at C2 (number of voids analyzed)	% Enrichment at C8 plus C5 (number of voids analyzed)
A	41.7	-0.03 ± 0.31 (28)	0.39 ± 0.17a (28)
B	41.7	-0.06 ± 0.35 (28)	0.19 ± 0.13a (28)

^aSignificantly greater than 0 by the Wilcoxon-paired sample test (P < 0.05).

To verify that our methods did not yield spurious positive enrichments, two paired-baseline values were subtracted from each other in subjects A and C (Table 1). In theory, this subtraction should have yielded 0% enrichment for all voids; however, this subtraction yielded non-significant mean % enrichments or mean % enrichments below 0 at C₂ due to unavoidable errors in the LC/MS/MS method and calculations [20]. The % ¹³C enrichments at C₈ in subject A was similar to the values that were found when two paired-baseline values were subtracted from each other, indicating no enrichment at C₈.

To test rhythmicity (circadian rhythm), the “run test” was applied to the data of subjects A and B [23]. As shown in Fig. 2, % ¹³C enrichment at C₂ had much fewer runs (i.e., 8) than the predicted number of 25 and 23 for these subjects, indicating rhythmicity in the data (P < 0.01). In contrast, % ¹³C enrichment at C₈ had 19 and 17 runs for these two subjects, which is consistent with a random pattern, indicating no rhythmicity (P > 0.05) [23]. Subject C did not have a sufficient number of voids to perform the test for rhythmicity.

3.2. [2-¹³C]glycine dose

There was low ¹³C enrichment at C₂ after the [2-¹³C]glycine dose (Table 1, Fig. 2). Of 28 voids collected for each subject, over 80% was 0 or negative ¹³C enrichment at C₂, indicating that the ratio of C₂/C₈ (plus C₅) after a [2-¹³C]glycine dose was low. In both subjects, significant ¹³C enrichments at C₂ after [2-¹³C]glycine were detected in only several voids that correspond to the timing of high C₂ enrichments from [¹³C]formate (Fig. 2). [2-¹³C]glycine enriched the C₈ plus C₅(red column) positions, and peak % ¹³C enrichments at C₈ plus C₅ were 0.62 and 0.42% in subjects A and B, respectively (Fig. 2, Table 1). Unlike the formate dose, no rhythmicity was found in the ¹³C enrichment at C₈ plus C₅.

4. Discussion

We found that [¹³C]formate predominantly enriched the C₂ position of purines, whereas significantly greater than zero ¹³C enrichment was only found at the combination of the C₈ and C₅ positions after a [2-¹³C]glycine dose (Table 1, Fig. 2). We shall describe our hypotheses to explain these findings below; however, we realize that these hypotheses (see Fig. 3) may require reevaluation or modification in the future, or they may be incorrect.

Fig. 3.

Proposed metabolic pathways explaining ¹³C enrichment of purine by [¹³C]formate and [2-¹³C]glycine. Using liver mitochondrial GCS and SHMT, ¹³C of [2-¹³C]glycine produces [3-¹³C]serine, which is transported to cytoplasm. Cytoplasmic SHMT and the trifunctional enzyme produce 5,10-H[¹³C]O-H₄folate, which is incorporated into the formyl moiety of formyl-GAR by GAR transformylase. [2-¹³C]glycine is also incorporated into GAR as glycine. Formyl-GAR is converted to AICAR, which is exported to plasma as its riboside and taken up by erythrocytes. This is converted back to AICAR by adenosine kinase in erythrocytes. Erythrocytes utilize [¹³C]formate and H₄folate by means of 10-HCO-H₄folate synthetase and AICAR transformylase to synthesize IMP. Thus, C₂ of purine is labeled by [¹³C]formate and C₈ plus C₅ are labeled by [2-¹³C]glycine (see Fig. 1). The liver converts inosine that is exported from erythrocytes to AMP. This AMP is then returned to erythrocytes as shown in order to maintain ATP concentrations. The coordinated IMP and AMP metabolism in the liver and erythrocytes has been previously established [38, 39].

4.1. Mechanistic explanation of the enrichment at the C position by [¹³C]formate

Our finding differs from the previously reported C₂/C₈ ratio of about 1.0 in animals [6-8]. We postulate that the liver stops purine biosynthesis de novo at AICAR, because isolated mammalian hepatocytes do not metabolize AICAR to IMP or to any other purines [24-28]. This is consistent with the low capacity of the mammalian hepatocytes or liver slices to synthesize purines in vitro from radioactive formate or serine [29-31]. The liver cannot enrich C₂ from [¹³C]formate, because AICAR cannot be metabolized to IMP (Fig. 3). Therefore, to explain our data, we tried to identify cells that: a) predominantly utilize formate as a source of one carbons; b) have AICAR transformylase and 10-HCO-H₄folate synthetase; and c) have an external supply of AICAR. We identified erythrocytes as being a prime candidate that fulfills all these requirements because erythrocytes: a) contain double the amount of formate compared to plasma [3]; b) possess the above two key enzymes but neither GAR transformylase, GCS, nor serine hydroxymethyltransferase (SHMT) [11, 32-34]; and c) are exposed to AICAR-riboside, or its base in the circulation, since AICA is a normal constituent of human urine [35]. In fact, AICAR accumulates in erythrocytes in Lesch-Nyhan syndrome, some forms of gout, and a genetic defect of AICAR transformylase [36, 37]. Therefore, we hypothesized that the formation of IMP from AICAR with formate and H₄folate occurs in erythrocytes. The idea of human erythrocytes participating in purine biosynthesis in such a way is not novel. Bertino et al [32] suggested this over 40 years ago based on the findings of abundant activities of 10-HCO-H₄folate synthetase and AICAR transformylase. In fact, Lowy et al [33] and Wagner and Levitch [34] reported that the in vitro incubation of intact erythrocytes with [¹⁴C]formate and AICA-riboside lead to the formation of IMP, and that this metabolism required H₄folate, ATP and the formation of AICAR from AICA-riboside. The mass of human erythrocytes, almost equal to that of the liver, should not be underestimated in its ability to metabolize purines [38]. In human erythrocytes, however, the lack of GAR transformylase and most of the other enzymes participating in purine biosynthesis precludes the ¹³C enrichment at C₈ [33].

Let us explain our thoughts on how erythrocytes and the liver metabolize and shuttle purines using Fig. 3. Firstly, where does AICAR in erythrocytes come from? We propose that the liver synthesizes AICAR de novo and exports into the circulation as AICA-riboside, which is then phosphorylated to AICAR by erythrocyte adenosine kinase. It has been shown that isolated rat hepatocytes cycle AICAR to AICA-riboside and the latter can be released from these cells to the medium [28]. Secondly, what is the fate of erythrocyte IMP, a minor component of its purine pool? Other researchers have established that IMP cannot be converted to AMP in erythrocytes [33, 39]; therefore, it must be exported to the circulation as inosine or hypoxanthine [40]. It has been shown that human erythrocytes loaded with [8-¹⁴C]IMP ex vivo rapidly release radioactivity in vivo (t1/2 = 1 hour) when re-injected back to their donor [40]. Presumably IMP is released as inosine or hypoxanthine, that are present in blood at 0.1 - 0.5 μM [41]. It is known that AMP is again exported from the liver to the circulation as adenosine that is incorporated back to erythrocytes and metabolized back to AMP by adenosine kinase to maintain adequate erythrocyte ATP concentration [38, 39] (Fig. 3). Plasma adenosine concentrations range from 0.1 to 0.5 μM [41].

We now present possible reasons why [¹³C]formate failed to enrich C₈. This may be due to the dilution of 10-H[¹³C]O-H₄folate by channeling of 10-H[¹²C]O-H₄folate to hepatic GAR transformylase (Fig. 3). Avian hepatic GAR transformylase forms a complex with the trifunctional enzyme and SHMT [42, 43]. Thus, 5,10-CH₂-H₄folate or 5,10-CH=H₄folate, formed from glycine, serine and histidine, could be channeled to GAR transformylase as 10-H[¹²C]O-H₄folate resulting in the dilution of the ¹³C enrichment at C₈. Although this is a possible explanation, the existence of such an enzyme complex, however, has not been proven in human liver. In addition, formate is readily oxidized such that about 25% of a tracer dose of [¹⁴C]formate is lost as¹⁴CO₂ in 2 hour, whereas only 1% is incorporated into uric acid in 11 days in humans [19]. Some of this oxidation likely takes place in hepatic peroxisomes [44]. Thus, formate is potentially metabolized to CO₂ rather than participating in purine biosynthesis in human liver.

We observed the circadian rhythm in the ¹³C enrichment at C₂. Erythrocyte adenosine kinase activity parallels blood inosine and hypoxanthine concentrations with a similar circadian rhythm [41]. Therefore, it is possible that the circadian rhythm in adenosine kinase activity could account for rhythmicity in the ¹³C enrichment at C₂ (Fig. 3).

4.2. Mechanistic explanation of the enrichment of the C₂, C₈ and C₅ positions by [2-¹³C]glycine

We postulate that the formation of GAR and formyl-GAR from [2-¹³C]glycine occurs in the liver [45], and enriches both C₈ and C₅ (Fig. 1 and 3). We are forced to report in this way, because our method does not distinguish independent enrichment at C₈ and C₅ [20]. Based on the findings by Pimstone et al [46], this may not impose problems in interpreting our ¹³C enrichment data at C₈. They found that 20% of [2-¹⁴C]glycine is incorporated into the C₂ and C₈ positions through folate-dependent reactions and the remaining 80% into the C₄ and C₅ positions through the folate independent pathway [46]. They used a method involving degradation of uric acid that does not distinguish between the ¹⁴C incorporation at C₂ and C₈. Although their data cannot be directly compared to ours, we hypothesize that the majority of this 20% of the ¹³C enrichment found in our study is at C₈. It is unlikely that our method was not sensitive enough to detect ¹³C enrichment at the C₂ positions. Our data further agree with those by Heinrich and Wilson [7] who found that there was no labeling at C₂ of guanine in rat carcass after [1,2-¹⁴C]glycine administration, whereas C₈ was labeled.

The substantial ¹³C enrichments at C₂ after [2-¹³C]glycine in subject B corresponded to the timing of high enrichments at C₂ by [¹³C]formate (Fig. 2). This finding suggests that some [¹³C]formate was produced from [2-¹³C]glycine or 10-H[¹³C]O-H₄folate and enriched C₂. The [2-¹³C]glycine dose used by us might have made a small but detectable contribution to the [¹³C]formate pool. However, considering many metabolic pathways involving glycine and our larger dose of glycine than formate on molar basis, its pathway to formate may be minor. Various substrates contribute to the formate pool without involving folate coenzymes, including methylthioadenosine, tryptophan, choline, acetate, and others [47-49].

Relatively low ¹³C enrichment at C₂ with [2-¹³C]glycine supports our interpretation of the [¹³C]formate data because erythrocytes with the absence of mitochondria have no GCS activity, whereas the liver has high GCS activity [11]. A possible metabolic pathway for [2-¹³C]glycine includes the following. In hepatic mitochondria, GCS with H₄folate cleaves [2-¹²C]glycine to 5,10-[¹³C]H₂-H₄folate, CO₂ and NH₃. Mitochondrial SHMT in the presence of glycine converts 5,10-[¹³C]H₂-H₄folate to [3-¹³C]serine, which is then transported to the cytoplasm [2, 11] as shown in Fig. 3. Cytoplasmic 5,10-[¹³C]H₂-H₄folate is formed from H₄folate and [3-¹³C]serine catalyzed by cytoplasmic SHMT and further metabolized to 10-H[¹³C]O-H₄folate by trifunctional enzyme (Fig. 3).

The % C₁₃ enrichment at C₈ of 0.19 - 0.39 from [2-¹²C]glycine is less than that at C₂ of 0.40 - 0.84 from [¹³C]formate in subjects A and B, even though the glycine dose was greater (Table 1). This may be due to channeling of ¹²C and dilution of ¹³C by ¹²C in the GAR transformylase, trifunctional enzyme and SHMT complex as discussed above [42, 43].

Using [2-¹³C]glycine, we unavoidably tested the ¹³C enrichment as if we used [3-¹³C]serine. As we discussed previously, [2-¹²C]glycine and [2-¹³C]glycine can form [3-¹³C]serine in the presence of GCS and SHMT in hepatic mitochondria [11], and [3-¹³C]serine can be transported to the cytoplasm (Fig. 3) [2].

4.3. Conclusion

The ¹³C enrichment at C₂ of uric acid was greater than at C₈ after a [¹³C]formate dose, and a circadian rhythm was seen in the ¹³C enrichment at C₂ in humans. After a [2-¹³C]glycine dose, however, no significant ¹³C enrichment at C₂ was found. To our knowledge, this is the first study to measure ¹³C enrichment from ¹³C-labeled formate and glycine independently into the C₂ and C₈ positions of purine in the same subjects. Although the number of subjects was small, the specificity and consistency of our data are compelling. However, it is necessary to stress that this research is in its infancy, and further investigations are required to confirm our findings and to prove our hypotheses.

Contrary to popular belief, our data suggest that ¹³C from these sources behave differently, and the incorporation of ¹³C of formate and the second ¹³C of glycine into purines may require coordination, which could be made not only on the molecular level but also on the organ level (erythrocytes and the liver). We take it for granted that purine metabolism presented in textbooks obtained mostly using uricotelic animals and microorganisms can be extrapolated to humans; however, our data suggest how fragile and precarious such an assumption is. Our non-invasive method of oral [¹³C]formate or [2-¹³C]glycine dosing and urine collection, although expensive, could be used to better understand human purine biosynthesis.